There appear to be two separate arms originating from the same general location at the crust-mantle boundary. One branch slopes northeast to feed the Yellowstone caldera, while a second branches off toward the Snake River Plain. The branches split in a way that the volcano-free zone between the two features results.
The researchers reasoned that, whatever else was going on to provide molten material, the paths to the surface were likely to be enabled by stresses in the crust. And that was going to depend on both the existing features in the crust (obtained largely through seismic data) as well as larger-scale processes going on in the mantle underneath. So, the model included both basic geological details, known physical processes, and a bit of history in the sense of what we know about how that section of the crust came to be.
And that’s where we come back to the Farallon plate. Its remains, having been driven beneath the North American plate, are continuing to sink and move through the mantle. That, the researchers surmise, is driving a general eastward flow of material through the viscous mantle. Just east of Yellowstone, however, that flow runs into the older border of the North American plate, where the crust is thicker and denser than the portion of the continent that was put in place by the Farallon plate.
New pathways
This thick crust causes the flow of the mantle to dip downwards. And that change in flow causes a series of stresses in the crust itself, most notably a compressive force between the older and newer sections of the North American plate, as well as a downward drag on the older section. Adding to the local stresses is the fact that all the material that erupted to form the Snake River Plain is denser than much of the rock that surrounds it, which generates strain on nearby rocks as it tries to sink.